Volume preserving projection filters and continuation methods in topology optimization

• Performance of continuation methods together with local filter operators for topology optimization problems is studied.
• Volume preserving Heaviside filter is used for obtaining 0–1 discrete topologies.
• Modifications to dual sequential algorithms are presented to handle the increased non-convexity with Heaviside filter.
• A continuation scheme is presented that yield good results in terms of both 0/1 discreteness and optimal performance.

Performance of continuation methods together with local filter operators in context of the density-based formulations for topology optimization problems is studied. In order to obtain binary discrete topologies, the volume preserving Heaviside filter is used together with continuation schemes on material penalization coefficient and filter parameters. Modifications to dual sequential optimization algorithms are presented to handle the increased non-convexity with Heaviside projection filter. Various continuations schemes are studied on test cases that include large-scale minimum compliance and compliant mechanism design problems. Finally, a continuation scheme is presented that yields good results in terms of both binary discreteness and optimal performance.